Inbolt has introduced a new bin picking system that integrates real-time 3D vision directly onto a robot arm, aiming to improve automation in unstructured industrial environments. The company announced the launch in Detroit, stating that the solution is designed to allow robots to identify, grasp, and place parts even when items are randomly positioned or partially obscured.
The system differs from conventional bin picking setups, which typically rely on fixed overhead cameras, detailed calibration, and predefined grasp points. According to Inbolt, mounting a 3D camera on the robot arm enables continuous perception and adjustment during the picking process. The company reports that this approach can achieve pick cycle times of under one second and success rates of up to 95 percent in live automotive production.
Inbolt’s solution uses what it describes as a closed-loop process inspired by human handling of objects. The robot identifies a pickable surface, analyzes the object after it is grasped, and adjusts its trajectory during movement to ensure accurate placement. This process, referred to by the company as in-hand localization, is intended to compensate for part variability and bin movement without requiring precise initial positioning.
The on-arm camera architecture is also positioned as a way to reduce system complexity and cost. By eliminating the need for multiple fixed cameras and high-resolution overhead installations, the same robot can be redeployed across different bins and configurations. Inbolt states that this reduces hardware requirements and shortens setup times for manufacturers.
The company reports that the bin picking system is already in operation at more than five factories, where it has been used in production environments. The platform runs on NVIDIA hardware and uses Inbolt’s proprietary AI models for real-time pose estimation and trajectory correction. According to the company, this architecture is designed to balance computational efficiency with consistent performance across different part geometries.
Inbolt says manufacturers interested in the technology can request a demonstration or pilot deployment through its website. The company develops AI-based robot guidance systems that combine real-time 3D vision with proprietary models for industrial automation and states that its technology is deployed across more than 70 factories worldwide.
